| Summary: | 碩士 === 國立高雄應用科技大學 === 電機工程系碩士班 === 94 === Abstract
Under the architecture of ADSL, which serve the customers from the host to the Internet by way of different kind of the Operators’ Telecommunication equipments, such as ATU-R, DSLAM, ATM, etc. This increases not only the complexity of the management, but also the transmission delay and causes much bandwidth waste. The reason is because the new port allocation to the new apply customer is assigned by the order of the sequence of the customers apply for, not consider about the total traffic load of the system. Consequently, it is easy to lead to the unbalance traffic distribution and data congestion. According to the observation of the real traffic distribution from the terminal of network management, it finds out the unbalanced traffic distribution is due to the port installation order by the sequence of the customers apply for. This thesis proposes the approach, the optimal distribution allocation for the traffic load of Switch, to deal with this problem.
GA algorithm is a heuristic optimization method, it searches by the way of adaptively random-guess. The main GA scenario is firstly to set up one population which is composed of many candidate individuals. They have little differences between the candidate individuals. Then make them up by simulating the gene operation of the biological species, such as crossover, mutation, selection. It uses an adaptive function to calculate the temporary result for the decision, whether the evaluation of chromosome of the population meets the condition to come to an end, or needs to do further evolve under such evolving environment. This thesis proposes the GA algorithm to find out the solution of the balance traffic load for every interface in L3 Switch. And this thesis investigates the problem of how to optimize the distribution allocation for the traffic load of every interface in L3 Switch by using the real traffic data for CHT (Chunghwa Telecom). The results show the proposed GA algorithm can effectively solve the data congestion, and reach the balanced traffic distribution.
|
|---|
